Proteolytic Control of Lipid Metabolism

Abstract

Synthesis and regulation of lipid levels and identities is critical for a wide variety of cellular functions, including structural and morphological properties of organelles, energy storage, signaling, and stability and function of membrane proteins. Proteolytic cleavage events regulate and/or influence some of these lipid metabolic processes and as a result help modulate their pleiotropic cellular functions. Proteins involved in lipid regulation are proteolytically cleaved for the purpose of their relocalization, processing, turnover, and quality control, among others. The scope of this review includes proteolytic events governing cellular lipid dynamics. After an initial discussion of the classic example of sterol regulatory element-binding proteins, our focus will shift to the mitochondrion, where a range of proteolytic events are critical for normal mitochondrial phospholipid metabolism and enforcing quality control therein. Recently, mitochondrial phospholipid metabolic pathways have been implicated as important for the proliferative capacity of cancers. Thus, the assorted proteases that regulate, monitor, or influence the activity of proteins that are important for phospholipid metabolism represent attractive targets to be manipulated for research purposes and clinical applications.

Graphical abstract

One of the primary uses the cell has for lipids is as building blocks for its various biological membranes that enclose specialized compartments. This is accomplished with the formation of a bilayer of two leaflets composed of mostly amphipathic phospholipids (PLs) whose hydrophobic acyl chains face each other to aggregate because of the hydrophobic effect while the hydrophilic regions face outward toward aqueous environments.¹ PL shape, as determined by their headgroup identities, includes both bilayer- and non-bilayer-forming lipids. For example, phosphatidylcholine (PC) is bilayer-forming because of its cylindrical shape, whereas PLs such as cone-shaped phosphatidylethanolamine (PE) contribute to negative membrane curvature as a result of their smaller headgroup relative to their acyl tails. Lipids involved in producing the opposite curvature, positive curvature, such as lysophospholipids, which have only one acyl chain, or phosphatidylinositol phosphates (PIP, PIP2, PIP3), present an inverted cone shape.^2,3 The length and desaturation of acyl chains contribute to membrane fluidity in that longer acyl chains serve to increase the number of interactions between other acyl chains to cause rigidity and desaturations break up the stacking of acyl chains to increase fluidity. Cholesterol is another major type of lipid found in membranes that impacts fluidity and serves as a precursor to other sterols, such as hormones. The role of cholesterol in fluidity is complex, but generally, it fills gaps formed by desaturations and thus rigidifies membranes.⁴ Lipids are asymmetrically distributed both among different organellar membranes and even between two leaflets of a singular membrane. For example, in the plasma membrane, phosphatidylserine (PS) is found on the inner leaflet, and the only organelle to contain cardiolipin (CL) is the mitochondrion.^5–9 Their unequal distribution establishes independent roles for those PLs. In this context, PLs not only contribute to organelle morphology, but additionally, they help confer organelle identity, act as signaling molecules, and impact protein stability.

Proteolytic events are accomplished by peptide bond cleavage—many at protease—specific sites—along the target protein. The outcomes of protease activity include quality control and turnover of old or misfolded proteins and subsequent degradation, regulation of enzyme activity, and release of a tethered protein from a membrane for relocalization, among others. Some proteases are highly promiscuous, such as digestive enzymes, while others have high specificity to targets and execute precise cleavage events, such as the tobacco etch virus proteases.^10–12 Proteases can serve as regulatory switches for major cellular pathways by quickly turning on or off the activity of another protein. In this way, one method by which they can influence a protein’s physiological role is by determining their half-lives. Dysfunction of protease activity can have many deleterious downstream consequences both inside and outside of cells. In this review, we highlight the numerous roles that proteases play in regulating lipid metabolism, in particular those pathways that reside in the mitochondrion.

PROTEOLYTIC CONTROL OF CHOLESTEROL METABOLISM

Sterol Regulatory Element-Binding Proteins.

Sterol regulatory element-binding proteins (SREBPs) are an important subfamily of basic helix–loop–helix leucine zipper transcription factors that are evolutionary conserved from yeast to mammals. Prior to becoming activated, SREBPs are integral membrane proteins that are retained in the endoplasmic reticulum (ER). In response to specific stimuli (see below), membrane-bound inactive SREBPs are proteolytically processed, releasing a soluble domain that regulates the expression of more than 30 genes.¹³ The cohort of affected genes are involved in either stress adaptation or the synthesis of cellular building blocks such as fatty acids and cholesterol.^14,15

Cholesterol is an essential component of cell membranes that because of its extremely hydrophobic character is an important modifier of membrane thickness, fluidity, and permeability.¹⁶ It composes ~15% of cellular lipids and serves many important roles in whole-organism physiology by acting as a precursor for steroid hormones and bile acids. Bile acids, produced by the liver, are critical for the uptake of dietary fats, including cholesterol, and are recycled multiple times daily. Cells ingest exogenous cholesterol that is packaged into circulating lipoproteins following their recognition by specific receptors, such as the ubiquitously expressed low-density lipoprotein receptor (LDLR). In addition, cells can make cholesterol de novo from acetyl-CoA. When cellular cholesterol levels are low, SREBPs respond and activate the expression of genes that encode enzymes involved in cholesterol biosynthesis—including 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase, HMG-CoA synthase, and mevalonate kinase—as well as those that enable its uptake from exogenous sources, most notably LDLR.^17,18

Cholesterol Regulation of SREBPs in Mammals.

In mammals, two SREBP genes, SREBF1 and SREBF2, encode three proteins.¹³ The first, SREBF1, is alternatively spliced to produce two isoforms, SREBP-1a and SREBP-1c.¹⁹ SREBP-1c, which has a shorter N-terminal acidic domain, is a weaker transcriptional activator than SREBP-1a.²⁰ All three mammalian SREBPs, SREBP-1a, SREBP-1c, and SREBP-2, have an N-terminal transcription factor domain, a short hydrophobic region localized at the middle of the protein, and a C-terminal regulatory domain.²¹ The two termini reside in the cytosol, while a short loop is formed in the ER lumen.²² The three SREBPs play different roles in lipid synthesis. SREBP-1c is involved in the synthesis of fatty acids; SREBP-2 preferentially regulates cholesterol synthesis; and SREBP-1a can regulate both cholesterol and fatty acid synthesis.^23–26

SREBP activation involves a sensing step that motivates movement from the ER to the Golgi where proteases reside that release the active N-domain via proteolytic processing of SREBP (Figure 1A). In mammals, SREBP binds to SREBP cleavage-activating protein (SCAP), which has a sterol-sensing domain.²² In the presence of cholesterol, SCAP binds cholesterol and interacts with INSIGs, ER-resident proteins that retain the SCAP–SREBP complex in the ER.²⁷ In the absence of cholesterol, the conformation of SCAP changes such that it no longer interacts with INSIGs. The dissociated INSIG is then ubiquitinated by the E3 ligase GP78 and degraded by the proteasome.²⁸ In parallel, the released SCAP–SREBP complex traffics to the Golgi to gain access to proteases required for SREBP activation.

Figure 1.

Regulated proteolytic activation of SREBPs. (A) Sterol regulatory element binding proteins (SREBPs) are synthesized as inactive precursors and retained in the ER. In mammals, SCAP responds to low cholesterol levels by releasing INSIGs and escorts SREBP to the Golgi. Flies, which lack INSIG orthologues, allow fly SREBP to move to the Golgi in response to low levels of PE. Once in the Golgi, SREBP is sequentially cleaved by S1P and S2P to release nuclear SREBP (nSREBP). nSREBP moves to the nucleus, where it activates the transcription of SRE-containing genes. (B) In fission yeast, hypoxia triggers proteolytic activation of Sre1. When oxygen levels are low, Scp1 escorts Sre1 to the Golgi, where it is then cleaved by Rbd2 in a process that is facilitated by Cdc48. Following its cleavage by an unidentified second protease, Sre1N is released to the nucleus to activate transcription of hypoxic response genes.

In the Golgi, activation of SREBPs occurs by two sequential proteolytic cleavage events (Figure 1A). The first cleavage is performed by site-1 protease (S1P), which binds to the complex and cleaves the middle loop region of SREBP.²⁹ The second protease, site-2 protease (S2P), cleaves at the N-terminal side of the transmembrane domain, leading to the formation of nuclear SREBP (nSREBP).^22,29–31 Upon moving into the nucleus by virtue of a nuclear localization signal³² nSREBP binds to sterol response elements (SREs) to activate target genes. SCAP is then recycled back to the ER, where it can form another SCAP–SREBP complex. After nSREBP has stimulated the transcription of genes involved in cholesterol and fatty acid synthesis, it is ultimately degraded by the ubiquitin–proteasome pathway.^33,34

Phosphatidylethanolamine Regulation of SREBPs in Drosophila.

In Drosophila, the predominant phospholipid is PE, whose abundance controls the activation of SREBP.¹⁵ While Drosophila has only a single SREBP gene, it has orthologues of mammalian SCAP, S1P, and S2P.³⁵ The proteolytic processing of SREBP is similar in mammals and Drosophila (Figure 1A). When PE is limiting, fly SCAP releases fly SREBP for transport from the ER to the Golgi, which allows the generation of nSREBP following the sequential action of the two Drosophila proteases, dS1P and dS2P.³⁶ Nuclear SREBP enhances the transcription of genes encoding enzymes of fatty acid biosynthesis.^15,36

Oxygen Regulation of SREBPs in Schizosaccharomyces pombe.

While cholesterol and PE levels are the main determinants of SREBP activation in mammals and flies, in the fission yeast Schizosaccharomyces pombe, this pathway is activated by hypoxia^37,38 (Figure 1B). Oxygen is required for sterol synthesis, and sterol levels decrease when oxygen is limiting. Under hypoxic conditions, Scp1, the yeast SCAP homologue, initiates the transport of Sre1 (yeast SREBP) from the ER to the Golgi.³⁷ In the Golgi, Sre1 is cleaved by the Golgi rhomboid protease Rbd2³⁹ in a process that is facilitated by the AAA-ATPase Cdc48.⁴⁰ Following its subsequent cleavage by a second currently unidentified protease, the functional Sre1, Sre1N, leaves the Golgi for the nucleus to activate transcription of genes required for hypoxic adaptation. When oxygen becomes available, newly synthesized Sre1 is retained in the ER by binding Scp1, and the oxygen-sensing Ofd1 promotes proteasomal degradation of Sre1N.⁴¹ Thus, while the basic SREBP regulatory scheme is evolutionarily conserved, the triggers for SREBP proteolytic activation, the proteases that release the soluble N-terminal SREBP domain, and the downstream targets of the released transcription factors are not.

ROLE OF PROTEOLYTIC CLEAVAGE IN MITOCHONDRIAL PHOSPHOLIPID METABOLISM

Overview of Mitochondrial Phospholipid Metabolism.

Mitochondria play crucial roles in cell survival and health. Their morphology is characterized by a double-membrane system, the inner membrane (IM) and outer membrane (OM). The aqueous environment between these two membranes is the intermembrane space (IMS) and the inside contained by the IM forms the matrix. Mitochondria have numerous functions, including the generation of indispensable energy in the form of ATP through oxidative phosphorylation (OXPHOS), and are dependent on lipid homeostasis for healthy and efficient ATP production^42,43

Mitochondrial membranes require PLs, which serve as both direct building blocks and precursors for other PLs that are made in this organelle. Lipids fated for the IM, though containing hydrophobic regions, have to be fluxed across the aqueous IMS compartment, and lipids synthesized in the IM needed elsewhere in the cell have to be trafficked in the reverse direction. These processes are thought to be aided by both lipid transfer proteins and membrane contact sites.^44–49 The mitochondrial contact site and cristae organizing system (MICOS) plays a role in forming cristae junctions and maintaining morphology as well as bringing the two mitochondrial membranes into close opposition.^46,50–54 It is possible that this supports lipid transfer by reducing the distance traversed in the aqueous compartment or perhaps by directly forming a conduit that increases the activity of lipid transport proteins. Likewise, in yeast contact sites between the OM and ER are mediated by the ER–mitochondria encounter structure (ERMES) complex to aid in lipid transfer.^44,55,56 While the ERMES complex is not conserved in higher eukaryotes, several other putative OM–ER bridges have been implicated as facilitating lipid exchange between these two organelles.^57–60 The interfacial region of lipid bilayers, which is defined as the boundary between hydrophilic headgroups and hydrophobic acyl chains, has the propensity to destabilize the folded state of proteins.^61,62 This poses a unique challenge for proteins involved in lipid flux and metabolism because they perform their functions in this intrinsically hostile environment.

Next, the metabolism and functional roles of three major constituents of mitochondrial membranes are discussed in greater detail, as all three are influenced in some capacity by the activity of mitochondrial proteases.

Phosphatidylethanolamine.

Typically the second most abundant PL in the cell, PE is a cone-shaped PL that contributes to membrane curvature,^63–65 is involved in membrane fission and fusion,^66,67 serves as a precursor for other PLs as well as GPI anchors,^68,69 and is important for cellular respiration.^42,70 PE can be produced by four distinct pathways. There is a mitochondrial pathway for PE production that is fed by ER-sourced PS.^71–73 In mitochondria, PS is decarboxylated by an evolutionarily conserved PL metabolizing enzyme. Encoded by the nuclear genome, phosphatidylserine decarboxylase 1 (Psd1 in yeast, PISD in mammals; Table 1 lists the yeast and mammalian names for lipid-related proteins discussed in this review) is imported into mitochondria, and following a self-cleaving autocatalytic processing event, it is anchored to the IM with its active site facing the IMS.^63,74–78 In yeast, Psd1 access to PS is granted by the activity of the lipid trafficking proteins Ups2/Mdm35 and may be facilitated by the MICOS.^47,79–81 Following the decarboxylation of PS, the newly produced PE can later be converted to PC once it is fluxed back out of the mitochondrion to the ER.^82–84 Psd1 is one of two major PE-producing pathways in yeast and is capable of generating the majority of cellular PE and PC.⁸⁵ When the mammalian orthologue, PISD, is knocked out in mice, it results in embryonic lethality.⁸⁶ PE is of broad physiological importance, and its deficiency has been linked to pathologies such as Alzheimer’s, Parkinson’s, and liver diseases.^87–89 Psd1/PISD is important for cell growth, OXPHOS, and mitochondrial morphology, and mutations in the human gene were recently shown to cause mitochondrial disease.^{42,70,90–93} With respect to OXPHOS in yeast, PE is important for complex IV activity, and PE made specifically in the IM by Psd1 is of particular importance for complex III function.⁹⁴ Muscle-specific ablation of PISD in adult mice causes a robust reduction in the O₂ consumption rate, ATP production, activity of OXPHOS complexes I–IV, and formation of respiratory supercomplexes while increasing oxidative stress.⁹⁵ Moreover, decreased PE is observed in muscle wasting, demonstrating the relevance of PE metabolism regulation in human health.⁹⁵

Table 1.

Mitochondrial Lipid-Metabolism-Related Proteins

name
in yeast	in mammals	description	localization
Mdm35	TRIAP1	obligate binding partner of Ups1 and Ups2; transports PA or PS across the IMS	IMS
Pgs1	PGS1	phosphatidylglycerophosphate synthase; catalyzes the committed step in CL biosynthesis	matrix
Psd1	PISD	Phosphatidylserine decarboxylase; generates PE from PS; active site faces the IMS.	IM
	STARD7	transports PC to the OM and from the OM to the IM; localization and function controlled by the context of PARL-mediated proteolysis	IMS and OM
Taz1	TAZ	remodels the acyl chain composition of CL; lyso-lipid transacylase; mutations cause Barth syndrome	IM and OM
Ups1	PRELID1	Ups1/Mdm35 heterodimer transports PA from the OM to the IM; transported PA fuels the CL biosynthetic pathway; requires Mdm35 for stability	IMS
Ups2	PRELID3b	Ups2/Mdm35 heterodimer transports PS from the OM to the IM; transported PS is converted to PE by Psd1/PISD; requires Mdm35 for stability	IMS

Phosphatidylcholine.

The most abundant cellular PL, PC is cylindrical-shaped and relatively enriched on the outer leaflet of the plasma membrane versus the inner leaflet.^96–99 As a bilayer-forming PL, PC contributes to organellar shape, and its biosynthesis occurs in the ER through either the CDP–choline pathway or by trimethylation of PE.^82,100,101 Thus, mitochondria acquire PC from an external source; this is thought to occur mainly through lipid transfer at ER–mitochondria contact sites, such as those established by ERMES in yeast.^44,55,56 In higher eukaryotes, this may be accomplished in conjunction with certain members of the STARD class of proteins,^102,103 as discussed further below. The conversion of PE to PC by PE-methyltransferase occurs mainly in the liver,¹⁰⁴ where PC is notably important for the stability and formation of lipoproteins.¹⁰⁵ When this pathway is deleted, mice develop hepatic steatosis and elaborate abnormal choline metabolites even when supplemented with choline, which should fuel the CDP–choline pathway.¹⁰⁶ Interestingly, in mammalian cells, high levels of PC inhibit SREBP-1 activity, and low levels result in increased mature SREBP-1 activity that accumulates in the nucleus.¹⁰⁷ In PC-deficient yeast lacking the PE to PC methylation pathway, the IM translocase for precursors that display an N-terminal mitochondrial targeting sequence, TIM23, is destabilized, resulting in decreased protein import; however, the OXPHOS machinery is unaffected.¹⁰⁸ The absence of the PC carrier STARD7 in murine HEPA-1 cells impairs OXPHOS activity and assembly, alters cristae organization, and reduces the cellular growth rate,¹⁰⁹ highlighting the importance of PC for mitochondrial function.

Cardiolipin.

Cardiolipin (CL) is a PL exclusively found in mitochondria in eukaryotic cells, but in support of the endosymbiotic theory, it is also found in bacteria. It is uniquely characterized by four acyl chains, unlike most phospholipids, which contain only two. CL is critical to the IM, making up 10–20% of the total IM lipid composition, and it is necessary for cristae morphology and the stability and function of OXPHOS complexes as well as mitochondrial solute carriers.^{65,70,110–113} CL has also been suggested to trap protons pumped into the IMS, which are concentrated on the IM surface and shuttled toward utilization by ATP synthase.¹¹⁴ CL biosynthesis begins with phosphatidic acid (PA), which is converted to phosphatidylglycerol (PG) through intermediary steps where PA is first converted to cytidine diphosphate diacylglycerol (CDP-DAG), then to PG-phosphate, and finally to PG. Lastly, one molecule of CDP-DAG is added to one molecule of PG to synthesize CL via CL synthase. After synthesis, the acyl chain composition of CL is remodeled by the combined actions of a CL lipase, identified as Cld1 in yeast,^115,116 and the conserved lysophospholipid transacylase taffazin (Taz1 in yeast, TAZ in mammals).^117–119 The end result of tafazzin-based remodeling is the generation of an enriched final form of CL with an increased content of unsaturated acyl chains.^120–123 In metazoans, CL can also be remodeled by acyl-CoA:lysocardiolipin acyltransferase-1 and monolyso-CL acyltransferase 1 to produce ulterior forms of CL by transferring various unsaturated acyl chains conjugated to coenzyme A onto monolyso-CL.^124–127 An important function provided by CL is related to its ability to facilitate the macromolecular assembly of numerous protein complexes. In addition to its well-established role in respiratory supercomplex formation,^43,128 CL is also important for the proper assembly of the MICOS, multiple dehydrogenase complexes with important metabolic functions, and the association of the m-AAA protease with prohibitins.^129,130 CL dysregulation is implicated in many illnesses, most notably the X-linked childhood cardiomyopathy known as Barth syndrome.^131–134 This disease results from mutations in the gene encoding tafazzin, leading to lower levels of CL of altered acyl chain composition and the accumulation of monolyso-CL, the remodeling intermediate.^132,133,135 These alterations in mitochondrial CL cause respiratory defects and irregular mitochondrial morphology, manifesting in the patient as muscle weakness, impaired growth, weakened immune system linked to cyclic deficits in neutrophils, cardiomyopathy, metabolic disorders, and other symptoms.^136–140

The Numerous Roles of Mitochondrial Proteases.

Within the mitochondrion, proteases exhibit a multitude of diverse functions (Figure 2). The IMS and matrix contain ATPases associated with diverse cellular activities (AAA) quality control (QC) proteases, yeast Yme1 (YME1L in higher eukaryotes) and Yta10/12 (AFG3L1/2 and SPG7 in higher eukaryotes), respectively (Table 2 lists yeast and human names for proteases discussed in this review).^141–144 These proteases have opposite topologies in the IM—the Yme1/YME1L containing i-AAA protease faces the IMS and the homo-oligomeric (AFG3L1/2) or hetero-oligomeric (Yta10/12, AFG3L1/SPG7, and AFG3L2/SPG7) m-AAA protease is exposed to the matrix—and degrade uncomplexed, misfolded, and old proteins to maintain organellar health as well as demonstrating other roles. The matrix also houses another ATP-dependent QC protease known as Pim1/LONP1, a member of the Lon family of serine peptidases.^145,146 Proteases in the OM have not been identified, and QC there is monitored instead by the ubiquitin proteasome system.^147–150 In addition to their important role in removing damaged proteins, mitochondrial proteases also process proteins into functional fragments. For instance, the m-AAA protease acts on the ribosomal subunit MrpL32 that contains a large unstructured N-terminal region that must be removed prior to ribosome assembly.^151–153

Figure 2.

Localization of mitochondrial proteases. Yeast proteins have only the first letter capitalized, and mammalian proteins are all capitalized. Examples of notable substrates for each protease are listed in the gray box associated with each protease.

Table 2.

Mitochondrial Proteases

name
in yeast	in mammals	description	localization
Atp23	XRCC6BP1	metalloprotease of the IM; participates in degradation of Ups1	IMS
Icp55	XPNPEP3	aminopeptidase that removes unstable N-terminal residues exposed by MPP	matrix
Imp1; Imp2	IMMP1L; IMMP2L	form the IM peptidase complex; required for maturation of certain IMS proteins	IM
Mas1; Mas2	MPPα; MPPβ	subunits of heterodimeric mitochondrial processing peptidase (MPP); removes N-terminal mitochondrial targeting sequence	matrix
Oct1	MIPEP	mitochondrial intermediate peptidase; removes octapeptides from selected imported proteins with an unstable N-terminal residue exposed by MPP	matrix
Oma1	OMA1	metalloendopeptidase; in mammals, involved in stress-regulated cleavage of OPA1; in yeast, has a role in turnover of Psd1ts	IM
Pcp1	PARL	serine protease and rhomboid intramembrane protease family member; in yeast, cleaves Mgm1; in mammals, cleaves STARD7 in TIM23-dependent and -independent manners	IM
Pim1	LONP1	member of the AAA+ and serine peptidase families; important for QC in the matrix	matrix
Yme1	YME1L	catalytic subunit of i-AAA protease; degrades unfolded/misfolded mitochondrial proteins in the IMS; substrates include Ups1, Ups2, STARD7, certain BTHS mutant tafazzins, and Psd1ts	IM
Yta10; Yta12	AFG3L1; AFG3L2; SPG7	subunits of homo- or hetero-oligomeric m-AAA protease; mediate degradation of misfolded or unassembled proteins in the matrix	IM

The mitochondrial targeting sequences of nuclear encoded proteins destined for the matrix or IM often are proteolytically removed; this is performed by the matrix-residing and soluble mitochondrial processing peptidase (MPP).^154,155 If the newly exposed amino acid post-MPP cleavage is destabilizing, then, as shown in yeast, the unstable N-terminus is removed in the matrix by either the IM-peripheral aminopeptidase Icp55^156–157 or the soluble mitochondrial intermediate peptidase Oct1.^158,159 IM protease 1/2 (Imp1/2 in yeast, IMMP1L/2L in mammals) is required for maturation of certain IMS proteins and contains two catalytic subunits differing in specificity.^160,161 Another protease in the IMS is Atp23/XRCC6BP1, which exemplifies the pleiotropic biological roles performed by mitochondrial proteases. Atp23 is a conserved metallopeptidase that in yeast removes the presequence from an mtDNA-encoded subunit of the ATP synthase, has chaperone activity essential for ATP synthase assembly, and participates in the degradation of Ups1, as discussed more below.^48,162,163

Pcp1 (PARL in mammals) is an IM-localized rhomboid protease, one of a family of intramembrane proteases that cleave proteins within membrane spanning segments and share a conserved core of six transmembrane helices forming a catalytic dyad. In yeast, Pcp1 generates short forms of Mgm1, which together with Mgm1 proteins that are not processed by Pcp1 helps mediate IM fusion.^164–166 It has been reported that mammalian PARL is capable of processing both pro-apoptotic Smac/DIABLO as well as the GTPase OPA1 (equivalent to yeast Mgm1), where the short form of OPA1 then inhibits apoptosis by slowing cytochrome c release that occurs subsequent to cristae remodeling.^167–169 However, it should be noted that the role of PARL in OPA1 processing is debated.¹⁷⁰ Intriguingly, OPA1, which normally is present in multiple forms that are collectively important for cristae morphology and IM fusion, is the substrate of two additional mitochondrial proteases besides PARL. During OPA1 biogenesis, YME1L mediates OPA1 processing and thus regulates the balance of expressed OPA1 products, though it is unclear whether this is accomplished in concert with another protease.^171,172 In response to stress in higher eukaryotes, OMA1, an IM metalloendopeptidase, is activated to cleave long OPA1 isoforms, which both promotes cytochrome c release to induce apoptosis and inhibits mitochondrial fusion.^173–176 In contrast, when the demand for mitochondrial energy is high, YME1L-mediated cleavage of OPA1 promotes mitochondrial fusion.¹⁷⁷ Therefore, while OPA1 is a substrate of both YME1L and OMA1, the consequences of these cleavage events are very different. Finally, to make matters even more interesting, upon depolarization of the IM, YME1L and OMA1 become substrates of each other, with the proteolytic victim determined by the availability of ATP.^178,179 Given the myriad of roles played by mitochondrial proteases, it is perhaps not surprising that they have emerging multifaceted roles in lipid metabolism.

Turnover of Lipid Transfer Proteins.

As previously mentioned, amphipathic PLs must traverse aqueous compartments between membranes of the ER and mitochondrion as a part of their biosynthetic pathways and to carry out their cellular functions. Complicating matters is the fact that mitochondria are not part of the secretory pathway. Lipid carrier proteins provide one means by which this challenge is overcome. Yeast lipid carriers Ups1 and Ups2 are related to the higher eukaryotic PRELI/MSF1 protein family (Table 1).^48,180 Ups1 (PRELID1 in mammals) is involved in trafficking PA from the mitochondrial OM to the mitochondrial IM, where PA is a precursor to CL synthesis.^{45,47,49,78,181} Thus, Ups1/PRELID1 is crucial for regulating CL levels (Figure 3). Ups2 (PRELID3b in mammals) is involved in trafficking PS from the OM to the IM, where PS is a precursor required for PE synthesis mediated by Psd1/PISD.^80,182,183 Thus, Ups2/PRELID3b is crucial for regulating PE levels, analogous to the role of Ups1/PRELID1 in CL metabolism (Figure 3). Both Ups1 and Ups2 must complex with Mdm35 (TRIAP1 in higher eukaryotes) to successfully accomplish their lipid transfer roles, and the structures of these complexes have been solved, providing further insight into their mechanisms.^{47,48,183–186} In yeast, there exists a final Ups homologue as a result of UPS2 gene duplication, Ups3, which also complexes with Mdm35; however, its exact function in PL transfer currently remains enigmatic.¹⁸⁴

Figure 3.

Proteolytic regulation of lipid transfer proteins in yeast. When CL levels are low, Ups1/Mdm35 transfers PA across the IMS to be converted to CL by a multistep biosynthetic pathway. When CL levels are high, Ups1/Mdm35 is sequestered at the IM by CL and Ups1 is degraded by Yme1 and Atp23. These proteases also degrade uncomplexed Ups1. The dynamic association of Ups1/Mdm35 is important for PA transport, and it is unclear after how many rounds of lipid transfer Ups1 dissociates from Mdm35 and is degraded. Similarly, Ups2/Mdm35 transfers PS across the IMS to be converted to PE by Psd1. Uncomplexed Ups2 is also degraded by Yme1, and the number of lipid transfer events of which the complex is capable has also yet to be determined. CL sequestration of Ups1/Mdm35 at the IM has yet to be formally tested in vivo.

It is notable that yeast Ups1 and Ups2 are intrinsically unstable compared with most mitochondrial proteins,⁴⁸ which typically have especially long half-lives. When complexed with Mdm35, Ups1 and Ups2 become more resistant to QC proteases such as Yme1, but not completely.^48,186 A hexamer of Yme1 subunits forms the i-AAA protease anchored in the IM that is tasked with monitoring the quality control of proteins exposed to the IMS.^{141,187–189} Because Ups1 and Ups2 are capable of transporting PL only when they are associated with Mdm35, constitutive turnover via proteolytic events of uncomplexed Ups1 and Ups2 proteins controls the flux; of PL precursors across the compartment by decreasing the availability of Ups1 and Ups2 for Mdm35 (likewise for PRELID1/PRELID3b and TRIAP1). There is potentially competition between Ups1 and Ups2 for Mdm35 because in Δups1 yeast, physiological levels of CL are restored when Ups2 is also lacking, while overexpression of Ups2 in wild-type yeast reduces CL levels.^45,190 It is thus conceivable that the association of Ups1 and Ups2 with Mdm35 could be regulated in order to adjust the relative levels of CL and PE within mitochondria.

The yeast Ups1/Mdm35 complex is capable of mediating the transport of PA in a bidirectional manner, at least in vitro. While Mdm35 is required for lipid transport by Ups1 and Ups2, its ability to dynamically dissociate and associate with Ups1 and Ups2 is critical for all steps of these transport processes, including lipid extraction, membrane binding, and lipid release.¹⁸⁵ It is intriguing to speculate that this dynamic behavior is a byproduct of the denaturing interfacial membrane region, which in this case has been productively harnessed. Interestingly, the presence of physiological levels of CL at the IM in in vitro studies increases the membrane association of the yeast Ups1/Mdm35 complex but decreases its delivery of PA to donor membranes.^47,186 Since CL-bound yeast Ups1 is subsequently degraded by Yme1 and/or Atp23, this suggests the existence of a feedback loop that matches the supply of lipid precursor with the demand for lipid product^48,186 (Figure 3). While capable of bidirectional flux in model systems, it remains to be determined whether an individual Ups/PRELID protein is capable of multiple rounds of lipid transport in vivo. Impaired CL accumulation in the IM due to loss of lipid carrier function in mammalian cells leads to release of cytochrome c, signaling apoptosis.^49,191,192 Human TRIAP1, the Mdm35 orthologue, is involved in cell-cycle progression through inhibition of p21 and is upregulated in multiple myeloma.^193,194 During low-level DNA damage, expression of TRIAP1 is increased by the transcription factor p53, augmenting cell survival. This tentatively links p53-mediated cell survival to mitochondrial CL and/or PE metabolism.^49,195 Manipulating this apoptotic pathway could provide one possible target in cancer treatment.

Proteolysis Determines Dual Localization of the Mammalian Lipid Transfer Protein STARD7.

Found in mammals, members of the STARD protein family are defined by their ~210 amino acid START domain, which is believed to be responsible for binding a single lipid; beyond that, however, the functions of many STARD family members are poorly understood. The START domain, structurally similar to the yeast Ups1/Mdm35 heterodimer, is found in 15 mammalian proteins, forms a hydrophobic tunnel responsible for lipid binding, and is also thought to potentially act as a lipid sensor for signaling.^{102,185,196,197} In humans, STARD1, originally termed steroidogenic acute regulatory protein (STAR), mediates the transfer of cholesterol from the OM to the IM, where it is cleaved to afford pregnenolone.¹⁹⁸ Though STARD1 demonstrates its activity at the OM, it contains a mitochondrial targeting sequence for matrix localization.¹⁹⁹ Potentially, this targeting allows excess STARD1, which might damage organelles, to deliver cholesterol to the IM^199,200 on its way to the matrix, where it is subsequently degraded.²⁰¹ Similarly, human STARD3/MLN64 mediates cholesterol transport from the ER to endosomes.²⁰² STARD4, -5, and -6 are soluble, lacking a subcellular localization motif, but are also thought to have a role in sterol transport.²⁰³ Interestingly, increases in SREBP levels result in increased levels of STARD4 in mouse liver and vice versa.^204–206 Of particular interest for this review, STARD7 is responsible for PC transfer either between the IM and OM or between the membranes exposed to the cytosol and the OM surface, depending on regulation via proteolytic cleavage.²⁰⁷ STARD7, along with STARD2 (also known as PC transfer protein, PCTP) and STARD10, is part of a subfamily of START proteins that are all capable of binding PC. STARD2 transfers PC between various cellular membranes, and STARD10 preferentially selects PC or PE containing a palmitoyl or stearoyl chain at the sn-1 position and an unsaturated fatty acyl chain (18:1 or 18:2) at the sn-2 position.^102,103 STARD11/GPBP traffics ceramides and DAG throughout the cell. The remaining START protein family members that have been studied are believed to function as GTPase activating proteins, lipase activators, and thioesterases.¹⁰²

In mammals, STARD7 is necessary for the accumulation of PC in the IM and for maintenance of respiration and cristae morphogenesis.¹⁰⁹ Interestingly, CL accumulation is impaired in the absence of STARD7, perhaps because PC can serve as an acyl chain donor for CL remodeling by tafazzin.^207,208 STARD7 mediates the flux of PC to both the mitochondrial OM and IM. To perform both tasks, STARD7 is dually localized to the cytosol and IMS. The final localization of STARD7 between these two compartments is dictated by the context in which it is cleaved by the transmembrane rhomboid protease PARL²⁰⁷ (Figure 4). For newly imported STARD7—which has entered the IMS through the translocon of the outer membrane (TOM)—PARL cleavage partitions the STARD7 precursor between two different mitochondrial import pathways, resulting in two outcomes: residence in the mitochondrial IMS or retro-translocation to the cytosol. The pathway toward which an individual STARD7 protein is funneled depends on whether TIM23 (the IM translocase for precursors that contain a mitochondrial targeting sequence) aids in the maturation process; also, it should be noted that these two import pathways occur simultaneously, not just one or the other. PARL-mediated cleavage of STARD7 during but not after mitochondrial import through the TOM promotes the partitioning of mature STARD7 to the cytosol.²⁰⁷ Mitochondria cannot synthesize PC on their own, so it is thought that in the cytosol, or possibly as anchored in the OM,²⁰⁹ STARD7 transfers PC to the mitochondrial surface from other membrane-enclosed compartments.^209,210 Alternatively, if the STARD7 precursor is threaded into TIM23, the translocon inserts STARD7 into the IM, stalling maturation by PARL until after completion of its import. Only after its lateral release from TIM23 into the IM does PARL then cleave STARD7 and release it into the IMS to become available to shuttle PC between the two mitochondrial membranes. The observation of TIM23-independent PARL-mediated cleavage was unexpected but clearly demonstrated through increases of cytosolic STARD7 when TIM23 was depleted.²⁰⁷ PARL thus preserves mitochondrial membrane homeostasis via STARD7 processing by regulating its localization.

Figure 4.

Dual targeting of the mammalian lipid-transfer protein STARD7 to the cytosol and mitochondrial IMS. STARD7 precursors are translated in the cytosol and threaded through the translocon of the OM (TOM), the common entry gate into mitochondria. If the precursor engages TIM23 in the IM, it is laterally sorted into this lipid bilayer prior to being cleaved by PARL. The released mature STARD7 is retained in the IMS, where it mediates the transport of PC across the IMS. If the STARD7 precursor is cleaved by PARL independent of TIM23, then it is released to the cytosol, where it facilitates the uptake of PC to the OM.

Prior to that study, STARD7 was only known to be localized at the OM and cytosol for PC transfer; thus, it was unclear at the time why STARD7 contained a mitochondrial targeting sequence directing it to the IM. Currently, it remains unknown why the stalling of STARD7 maturation via TIM23 causes it to be retained in the IMS if the mature forms of STARD7 after PARL cleavage are identical with or without TIM23. Such remaining important and unexplored questions will stimulate future investigations of the temporal aspects of STARD7 import and proteolytic cleavage. Potentially, TIM23-mediated import into the IM could help retain STARD7 in the IMS via a folding-trap mechanism in which the folding of the soluble lipid-binding domain prevents its retro-translocation through the TOM and into the cytosol. However, if TIM23 does not engage STARD7 and laterally insert it into the IM, then perhaps a significant fraction of STARD7 is still unfolded and/ or still being translocated through the TOM, such that once cleaved by PARL, it can slip out of the TOM complex back into the cytosol to fold (Figure 4). Like the previously discussed lipid carriers, STARD7 is also degraded by YME1L, which imposes upon it a shorter half-life than most mitochondrial proteins.²⁰⁷ Notably, mutations in STARD7 have been detected in asthma and dysfunctional lung epithelium phenotypes.^207,211,212 The mitochondrial dysfunction (which includes increased reactive oxygen species production, damaged mitochondrial DNA, decreased OXPHOS, and altered morphology) seen in STARD7-deficient cells’ though mechanistically unresolved, was associated with oxidant-mediated changes in epithelial barrier integrity and function.²¹²

DISEASE CONNECTIONS

A Yin–Yang Relationship between Mitochondrial Proteases and PL Metabolism.

A possible role for the i-AAA protease Yme1 in regulating mitochondrial PE metabolism was initially provided in a yeast screen that determined that the cellular and mitochondrial levels of PE and PC are altered in its absence.²¹³ Specifically, when Yme1 is missing, PE levels and Psd1 activity are both increased. On the basis of its documented QC-related activities, it was suggested that in yeast Yme1 may normally degrade Psd1 and that preventing this turnover increases the abundance of Psd1, which then increases the level of its lipid product.²¹³ One implication of this model is that Yme1 has the capacity to degrade functional Psd1, at least in yeast. This would suggest that Yme1 has a direct role in regulating mitochondrial PE metabolism beyond monitoring the QC of its assorted components. An alternative explanation for the observed changes in PE and PC in the absence of Yme1 reflects the role of this protease in degrading proteins that mediate the flux of PS into or PE out of the IM. As already mentioned, movement of PS into the IM across the IMS is mediated by Ups2/Mdm35 in yeast.^80,182 How PE moves across the IMS and whether Yme1 has a role in regulating this process have not been determined.

Recently, a direct role for Yme1 in the QC of Psd1 was established using a temperature-sensitive allele of yeast Psd1 (Psd1ts) that is unable to perform autocatalysis at the nonpermissive temperature.⁹¹ Interestingly, the proteases tasked with handling the turnover of misfolded Psd1ts vary depending on whether it has or has not performed autocatalysis. After autocatalysis, a shift to higher temperature causes Psd1ts to form aggregates whose accumulation is prevented/reduced by Yme1.⁹¹ If newly imported Psd1ts is prevented from performing autocatalysis at elevated temperature, then Oma1 generates a set of proteolytic fragments that are subsequently degraded by Yme1.⁹¹ It remains unresolved why the proteases involved in the turnover of Psd1ts differ depending on its autocatalytic status, especially since Psd1 is presumably anchored to the IM in the same manner both before and after autocatalysis.

One of the four mutations present in Psd1ts, K356, occurs two amino acids upstream of a conserved arginine, R358,²¹⁴ that was recently associated with a novel mitochondrial disease.⁹³ Moreover, the disease-associated amino acid occurs 13 amino acids downstream of a conserved histidine residue that is the base of the catalytic triad that mediates Psd1 autocatalysis.^91,215 On the basis of its proximity, it was speculated that the disease-associated R277Q mutation (equivalent to the R358Q mutation in yeast) may disrupt autocatalysis. Indeed, when modeled in either yeast Psd1 or human PISD, mutation of this conserved Arg disturbs autocatalysis, formally demonstrating that the R277Q mutation is pathogenic.⁹³ Interestingly, biochemical characterization of patient-derived fibroblasts suggests that the activities of certain IM proteases are impaired in the context of aberrant mitochondrial PE metabolism.⁹³ Thus, proteases are important for phospholipid metabolism and phospholipids are critical for the activity of proteases.

Mutations in mitochondrial proteases are also a cause of inherited human disease. For instance, a missense mutation in a functional domain within PARL has been recorded in a subset of Parkinson’s cases.²¹⁶ In addition, a missense mutation in YME1L that prevents its maturation in mitochondria, resulting in a short-lived protein, causes a recessive mitochondriopathy.²¹⁷ Disturbances in mitochondrial PL levels were not tested in either disease setting. While it is tempting to speculate that alterations in mitochondrial PL metabolism could contribute to the pathogenesis in these diseases, it is important to note that mitochondrial proteases, in addition to having additional substrates, also have central QC functions that are critical to the proteostatic fitness of this organelle. Thus, whether mitochondrial PL metabolism is impacted in the context of these patient-associated mutations in PARL or YME1L is presently unresolved.

Potential Therapeutic Target for a Subset of Barth Syndrome Patients.

Since it attaches acyl chains to lysolipids, tafazzin, the transacylase responsible for the acyl chain maturation of CL,²⁰⁸ must work within the denaturing interfacial region below the hydrophilic headgroups but above the hydrophobic core of the lipid bilayer. Taz1/TAZ, which is a monotopic protein whose membrane anchor extends into but not through the membrane, associates with mitochondrial membranes that face the IMS.²¹⁸ As mentioned previously, mutations in TAZ result in the childhood cardiomyopathy Barth syndrome (BTHS), the first known inherited disorder of mitochondrial phospholipid metabolism.^132,133,136 A series of studies were performed in which yeast was used as a model to determine whether disease-associated BTHS mutations are pathogenic, and if so, what the mechanistic basis for their lack of function is. A subset of four missense alleles were observed whose steady-state levels, which are normally low relative to wild-type Taz1, were restored in the absence of Yme1.²¹⁹ While the increased expression in the absence of Yme1 normalizes the assembly of all four BTHS mutants, it results in increased Taz1 activity only for two of these alleles. Additionally, even when Yme1-based degradation is not occurring, the mutant Taz1 complexes that form disassemble faster than normal and accumulate as aggregates.²¹⁹ This work provided yet another example of the intimate relationship between mitochondrial phospholipid metabolism and mitochondrial proteases. Additionally, since the loss-of-function mechanisms of the BTHS mutations that have been tested to date are conserved between yeast and humans,²²⁰ this study identified small molecules capable of stabilizing mutant TAZ complex assembly and/or preventing its recognition by YME1L as a potential therapeutic strategy for a subset of BTHS patients.

An Emerging Link between Mitochondrial Phospholipid Metabolism and Cancer.

An exciting link between mitochondrial phospholipid metabolism and cancer has recently emerged with the discovery of the novel tumor suppressor LACTB, which when overexpressed in breast cancer cells reduces cell proliferation and increases cellular differentiation via a mechanism that is at least in part explained by a significant decrease in the levels of PISD.²²¹ As the LACTB-mediated decrease in PISD abundance is post-transcriptionally mediated,²²¹ it is tempting to speculate that some of the QC proteases responsible for the turnover of yeast Psd1ts⁹¹ are harnessed by LACTB to ultimately decrease mitochondrial PE levels. In addition, LACTB can inhibit colorectal cancer; however, the ability of LACTB to suppress this type of cancer was apparently independent of any inhibition of mitochondrial PE metabolism via PISD.²²² Another link to cancer was recently established for TAZ, which was identified via a CRISPR screen to be required for acute myeloid leukemia (AML) cell growth and viability.²²³ In the absence of TAZ, AML differentiation is increased at the expense of its proliferative capacity, and in addition to the expected changes in the CL level, the PE and PS levels are decreased and increased, respectively. Surprisingly, increasing cellular PS either by supplementation, by deletion of PISD, or by inhibition of PISD activity pharmacologically had similar effects on AML growth and differentiation as observed when TAZ is missing.²²³

Cachexia, or the wasting syndrome accompanying severe chronic illness, is observed in 80% of cancers and is the direct cause of 20% of cancer deaths.^224,225 Interestingly, CL levels are elevated and the efficiency of ATP synthesis is reduced in cancer cachexia, raising the possibility that alterations in mitochondrial PL metabolism may play a role in the cachexia process.^226,227 In addition, the activity of one of the enzymes necessary for CL biosynthesis, PGP synthase, is increased in a cancer cell line harboring a mutation of STARD13.²²⁸ These results suggest that drugs designed to inhibit specific aspects of mitochondrial phospholipid metabolism, potentially by promoting the proteolytic degradation of key players, could have therapeutic value in treating certain types of cancer.

PERSPECTIVES FOR CHEMICAL BIOLOGY

In view of their key role in activating lipid metabolism, it is not surprising that compounds capable of inhibiting the generation of nSREBP have been actively sought with the twin goals of developing tools to better understand SREBP biology and therapeutic agents for human diseases, in particular cancer.

An important class of drugs that have been developed inhibit the translocation of the SCAP–SREBP complex from the ER to the Golgi. Specifically, fatostatin, betulin, and xanthohumol are inhibitors that block the activation of SREBP transcription factors. Fatostatin is a small synthetic diarylthiazole molecule that targets SCAP.^229,230 Betulin is also a small molecule, isolated from birch bark, that is a cell-permeable pentacyclic triterpenoid that interacts with SCAP.²³¹ Mechanistic studies of fatostatin and betulin indicate that they bind SCAP directly and prevent its dissociation from INSIG, which in the net suppresses the flux of SREBPs to the Golgi.^229,230 By preventing the activation of the SREBP transcriptional program, which in turn suppresses cell growth, these two inhibitors have been shown to prevent the proliferation of several types of cancer, including prostate, lung, pancreatic, breast, hepatocellular carcinoma, and glioblastoma.^231–239 Xanthohumol, a prenylated flavonoid from hops, is another SREBP inhibitor.^240,241 Similar to fatostatin and betulin, xanthohumol blocks ER-to-Golgi trafficking of the SCAP–SREBP complex; however, instead of targeting SCAP, it binds to Sec23/24 and prevents the incorporation of the SCAP–SREBP complex into COPII-coated vesicles.^240,241 Xanthohumol decreases the viability of many types of cancer, likely as a result of its ability to induce apoptosis.^242–244

There are also inhibitors that can prevent SREBP cleavage by inhibiting the Golgi-resident proteases S1P and S2P. PF-429242 is a potent inhibitor of S1P that downregulates SREBP activity.²⁴⁵ By preventing the proteolytic processing of SREBP, PF-429242 inhibits the generation of nSREBP, which in turn blocks the activation of cholesterol and fatty acid synthesis.^245,246 PF-429242 also induces apoptosis and inhibits cancer cell growth.^232,247 Nelfinavir, an HIV protease inhibitor, and 1,10-phenanthroline are S2P inhibitors that similarly prevent the release of nSREBP.^248–250 Like the other compounds targeting the SREBP pathway, these latter two inhibitors have anticancer properties, as they decrease cancer cell proliferation and increase apoptosis.^250,251

In contrast to the Golgi-resident proteases in the SREBP pathway, generally very little is known about the specificity of the mitochondrial proteases involved in the regulation of mitochondrial phospholipid metabolism. While some insight into the specificity of the i-AAA protease formed from YME1L has recently been established,^252,253 whether the identified substrates, which are components of the import apparatus that reside in the IMS, have identified basic principles that are relevant to the lipid metabolic pathways discussed in this review has not been determined. In the net, the absence of insight into the substrate specificity of the implicated proteases has hampered the identification of inhibitors. However, the recently solved structures of the i-AAA and m-AAA proteases^189,254 could facilitate efforts to rationally design inhibitors. Once developed/identified, inhibitors specific for individual proteases would represent invaluable tools to better understand their importance with respect to lipid metabolism. For example, such inhibitors would enable an interrogation of whether the constitutive turnover of the PA- and PS-transporting Ups1/PRELID1 and Ups2/PRELID3b is a consequence of the denaturing environment in which they must work or is instead an important aspect of the transport mechanism that is presently unappreciated. Another application for such compounds could be to dissect the potential importance of altered mitochondrial lipid metabolism in human diseases. For instance, it would be interesting to determine whether the mitochondrial dysfunction caused by a missense mutation in YME1L that results in a recessive mitochondriopathy²¹⁷ can be recapitulated with inhibitors, and if so, it would be useful to determine how its inhibition may impact mitochondrial lipids. Moving forward, pharmacological agents would enable new approaches that are amenable to kinetic analyses and thus greatly expand our mechanistic understanding of the myriad of mitochondrial proteases that modulate mitochondrial lipid metabolism.

The interaction of CL with the subunit of ATP synthase that forms the proton-conducting ring (the c subunit) was characterized with solid-state NMR spectroscopy.²⁵⁵ This analysis established that the building block of the rotating motor of ATP synthase interacts specifically with CL, an insight that is especially important since subunit c is a major drug target for diseases including tuberculosis and cancer.²⁵⁶ Thus, it is possible that this technique can be applied to study the interaction of CL and other lipids with complexes of interest. Such a strategy could be of particular value for better understanding the sequestration of yeast Ups1/Mdm35 by CL at the IM and evaluating the possibility that disease-associated mutations in TAZ, PARL, etc., could result in disturbances of their ability to productively interact with specific phospholipids. Compounds that disrupt the ability of lipid transport proteins to extract, bind, and/or release their lipid substrates would also represent new research tools that would provide a better understanding of the overall biology. For instance, how would such an inhibitor impact lipid transport protein half-life and could it be of potential therapeutic value by limiting lipid flux into the mitochondrion in a disease state?

One challenge intrinsic to lipid biology is the ability to biochemically monitor the distribution of phospholipids in distinct membrane compartments and surrounding different membrane proteins and protein complexes. We recently exploited the ability of styrene–maleic acid copolymers to generate lipid nanodiscs from native mitochondrial membranes that contain proteins and their surrounding lipids, termed SMA lipoprotein particles (SMALPs), to determine the lipid composition surrounding complex IV in the IM.⁹⁴ The generated results challenged PE trafficking dogma and established a new thermodynamic-equilibrium-based model of PE distribution between the IM and OM. This SMALP strategy, which can easily be expanded to survey additional complexes that occupy distinct membrane compartments, could prove useful in observing how the dysregulation of proteases or lipid carriers with the above-mentioned approaches affects the lipid environment of important mitochondrial complexes. Moreover, as discussed in another contribution in this Proteolytic Regulation of Cellular Physiology special issue,²⁵⁷ SMALPs could be utilized to isolate mitochondrial proteases in their native lipid environment in nanodiscs to preserve their activity and stability as well as structurally and functionally important lipid interactions. The protein-embedded nanodiscs could then be subjected to shotgun lipidomics, cryo-EM, and in vitro activity assays, among others. Pharmacological tools and chemical biology approaches will help unravel the complicated, multifaceted, and sometimes bidirectional relationship between proteases and lipid metabolism and may identify novel therapeutic targets and agents.

ABBREVIATIONS

AAAs

ATPases associated with diverse cellular activities

AML

acute myeloid leukemia

BTHS

Barth syndrome

CL

cardiolipin

ERMES

endoplasmic reticulum–mitochondria encounter structure

HMG-CoA

3-hydroxy-3-methylglutaryl-CoA

IM

inner membrane

IMS

intermembrane space

INSIG

insulin-induced gene

LACTB

serine β-lactamase-like protein

LDLR

low-density lipoprotein receptor

MICOS

mitochondrial contact site and cristae organizing system

MPP

mitochondrial processing peptidase

nSREBP

nuclear SREBP

OM

outer membrane

OXPHOS

oxidative phosphorylation

PA

phosphatidic acid

PC

phosphatidylcholine

PE

phosphatidylethanolamine

PG

phosphatidylglycerol

PIP

phosphatidylinositol phosphate

PL

phospholipids

PSD1

phosphatidylserine decarboxylase 1

QC

quality control

S1P

site-1 protease

S2P

site-2 protease

SCAP

SREBP cleavage-activating protein

SRE

sterol response element

SREBP

sterol regulatory element-binding protein

STAR

steroidogenic acute regulatory protein

STARD7

star-related lipid transfer protein 7

Taz/TAZ

taffazin

ts

temperature-sensitive

UP

ubiquitin proteasome

KEYWORDS

Barth Syndrome:

A rare multisystemic X-linked genetic disorder caused by mutations in the TAFAZZIN (TAZ) gene. Affected individuals have cardiomyopathy, skeletal muscle myopathy, cyclic neutropenia, and growth delay.

Cardiolipin:

A phospholipid distinct to the mitochondrial organelle in eukaryotes whose identity is atypical from other phospholipids in that it contains four acyl chains and two phosphate headgroups that are bridged by a glycerol. It is primarily located in the inner membrane of the mitochondrion and functions to support OXPHOS and mitochondrial morphology.

Sterol Regulatory Element-Binding Proteins:

Inactive membrane proteins that when released following proteolytic cleavage act as transcription factors for mostly cholesterol and fatty acid synthesis-related genes.

LACTB:

A mitochondrial protein with similarity to serine proteases of the bacterial β-lactamase superfamily that acts as a tumor suppressor capable of inhibiting the proliferation of breast cancer cells by decreasing PISD levels.

Lipid Transfer Proteins:

Proteins whose functions are to aid in the flux of amphipathic phospholipids across aqueous barriers to their membrane destinations.

Phosphatidylethanolamine:

An abundant phospholipid in most membrane compartments, phosphatidylethanolamine is a non-bilayer-forming lipid made by up to four different pathways that contributes to membrane curvature, fission and fusion events, and mitochondrial energy production.

Phospholipids:

Lipid molecules typically composed of two acyl chains linked through a glycerol moiety to a phosphate attached to a headgroup. Phospholipids are the building blocks of biological membranes.

Proteases:

Enzymes that cleave peptide bonds to break up proteins into smaller polypeptides with a multitude of downstream biological outcomes.

Quality Control:

Surveillance mechanisms that the cell employs to maintain cellular fitness. One portion of this system is enacted through the use of proteases that degrade misfolded, aggregated, damaged, or old proteins.